Tool Use

Tool use (also called function calling) is a capability that allows language models to call external tools — functions, APIs, databases, or services — as part of generating a response. Instead of relying solely on its training data, the model can access real-time information, perform calculations, execute code, search databases, and interact with external systems. The mechanism works by defining available tools (with their names, descriptions, and parameter schemas) in the API request. When the model determines that a tool would help answer a query, it generates a structured tool call (specifying which tool to use and what parameters to pass). The application executes the tool, returns the result, and the model incorporates the result into its response. Tool use is fundamental to AI agent architectures. While a basic chatbot can only respond with text from its training data, a model with tool use can check live inventory, create calendar events, send emails, query databases, and perform virtually any action that can be expressed as an API call.

Tool use transforms language models from passive text generators into active participants in business workflows. An AI assistant with tool access can look up a customer's order status, check product availability, schedule meetings, generate reports from live data, and execute multi-step business processes. This capability dramatically expands the range of tasks that AI can handle autonomously. Instead of providing information that a human must then act on, the AI can take actions directly — with appropriate guardrails and approval mechanisms in place. For organisations building AI-powered applications, tool use architecture is a key design decision. Teams must decide which tools to expose, how to handle permissions and authentication, what validation to apply to tool calls, and how to implement human-in-the-loop approval for high-stakes actions. Getting this architecture right is essential for building reliable, secure AI applications.